Perspective: AIAI Could Be a Force for Positive Social Change – but We’re Currently Heading for a Darker Future

Artificial Intelligence (AI) is already re-configuring the world in conspicuous ways. Data drives our global digital ecosystem, and AI technologies reveal patterns in data. Smartphones, smart homes, and smart cities influence how we live and interact, and AI systems are increasingly involved in recruitment decisions, medical diagnoses, and judicial verdicts. Whether this scenario is utopian or dystopian depends on your perspective.

Marcus Tomalin and Stefanie Ullmann write in The Conversation that the potential risks of AI are enumerated repeatedly. Killer robots and mass unemployment are common concerns, while some people even fear human extinction. More optimistic predictions claim that AI will add $15 trillion to the world economy by 2030, and eventually lead us to some kind of social nirvana.

They add:

We certainly need to consider the impact that such technologies are having on our societies. One important concern is that AI systems reinforce existing social biases – to damaging effect. Several notorious examples of this phenomenon have received widespread attention: state-of-the-art automated machine translation systems which produce sexist outputs, and image recognition systems which classify black people as gorillas.

These problems arise because such systems use mathematical models (such as neural networks) to identify patterns in large sets of training data. If that data is badly skewed in various ways, then its inherent biases will inevitably be learnt and reproduced by the trained systems. Biased autonomous technologies are problematic since they can potentially marginalize groups such as women, ethnic minorities, or the elderly, thereby compounding existing social imbalances.

If AI systems are trained on police arrests data, for example, then any conscious or unconscious biases manifest in the existing patterns of arrests would be replicated by a “predictive policing” AI system trained on that data. Recognizing the serious implications of this, various authoritative organizations have recently advised that all AI systems should be trained on unbiased data.